#!/usr/bin/env bash
# Hipocampus SessionStart hook
# 1. Auto-create memory directories and config if missing
# 2. Output memory protocol to stdout for agent injection

# ─── Resolve project cwd from hook stdin JSON ───

STDIN_DATA="$(cat)"
PROJECT_CWD="$(echo "$STDIN_DATA" | grep -o '"cwd":"[^"]*"' | head -1 | cut -d'"' -f4)"
if [ -n "$PROJECT_CWD" ] && [ -d "$PROJECT_CWD" ]; then
  cd "$PROJECT_CWD"
fi

# ─── Auto-setup (idempotent) ───

mkdir -p memory/daily memory/weekly memory/monthly knowledge plans 2>/dev/null

# ROOT.md stub
if [ ! -f memory/ROOT.md ]; then
  cat > memory/ROOT.md <<'ROOTMD'
---
type: root
status: tentative
last-updated: 1970-01-01
---

## Active Context (recent ~7 days)
<!-- Current work and priorities -->

## Recent Patterns
<!-- Cross-cutting insights -->

## Historical Summary
<!-- High-level timeline -->

## Topics Index
<!-- topic: keywords, references -->
ROOTMD
fi

# hipocampus.config.json
if [ ! -f hipocampus.config.json ]; then
  cat > hipocampus.config.json <<'CONFIG'
{
  "platform": "claude-code",
  "search": {
    "vector": true,
    "embedModel": "auto"
  },
  "compaction": {
    "rootMaxTokens": 3000,
    "cooldownHours": 3
  }
}
CONFIG
fi

# Layer 1 templates
[ ! -f SCRATCHPAD.md ] && echo "# Scratchpad\n\n<!-- Active working state -->" > SCRATCHPAD.md
[ ! -f WORKING.md ] && echo "# Working\n\n<!-- Current tasks in progress -->" > WORKING.md
[ ! -f TASK-QUEUE.md ] && echo "# Task Queue\n\n<!-- Queued items -->" > TASK-QUEUE.md

# ─── Output protocol to stdout ───

cat <<'PROTOCOL'
## Hipocampus — Memory Protocol (MANDATORY)

This project uses hipocampus 3-tier memory. The full protocol is in /hipocampus:core skill.
**All memory write operations MUST be dispatched to subagents** to keep the main session context clean.

### FIRST RESPONSE RULE — MANDATORY
**On the very first user message of every session**, before doing ANYTHING else:
Run the Session Start protocol below FIRST. This takes priority over ANY user request. Complete the step below, ONLY THEN respond to the user.

Read SCRATCHPAD.md, WORKING.md, TASK-QUEUE.md, memory/ROOT.md first.

**This procedure must be completed before responding to the user NO MATTER WHAT**
1. **DO NOT SKIP** **DO NOT COMPROMISE** **Compaction maintenance (cooldown-gated):**
   Read `memory/.compaction-state.json` and `hipocampus.config.json` (`compaction.cooldownHours`, default 3).
   - **Within cooldown:** Skip compaction subagent — no dispatch needed.
   - **Cooldown expired, file missing, or `cooldownHours` is 0:** Write `memory/.compaction-state.json` with `{ "lastCompactionRun": "<current ISO timestamp>" }`, then dispatch a subagent to run hipocampus:compaction skill (chain: Daily→Weekly→Monthly→Root), then run `hipocampus compact` + `qmd update` + `qmd embed`.
**This procedure must be completed before responding to the user NO MATTER WHAT**

### End-of-Task Checkpoint (mandatory — subagent)
After completing any task, **dispatch a subagent** to append a structured log to `memory/YYYY-MM-DD.md`.
Compose the log with ## headings per topic: what was requested, analysis, decisions with rationale, outcomes, files changed.
**You must provide the task summary to the subagent** — it has no access to the conversation.

### Rules
- **Never skip Session Start** — every session begins with it, no exceptions
- **Never skip checkpoints** — every task completion MUST append to daily log via subagent
- **All memory writes via subagent** — never pollute main session with memory operations
- memory/*.md (raw): permanent, never delete
- Search: use /hipocampus:search skill
- If this session ends NOW, the next session must be able to continue immediately
PROTOCOL